Josias-Ounsinli/Assessing-the-effects-of-various-vaccination-scenarios-on-malaria-dynamics

ASSESSING THE EFFECTS OF VARIOUS VACCINATION SCENARIOS ON MALARIA DYNAMICS IN AFRICA: A MACHINE LEARNING BASED APPROACH # Assessing Short and Long Term Malaria Transmission and Burden Under Various Vaccination Scenarios: A Machine Learning-Based Approach ## Abstract Malaria continues to be a highly life-threatening disease in sub-Saharan Africa. The 2022 World Malaria Report notes a slight increase in global cases, rising from 245 million in 2020 to approximately 247 million in 2021. In recent years, the World Health Organization (WHO) has earnestly addressed the fight against malaria. It has formally sanctioned the adoption and recommendation of two vaccines: RTS,S/ASO1 (Mosquirix®) and R21/Matrix-M™ (R21/MM). The rollout of vaccination programs is slated to commence in 2023, raising questions about the efficacy of vaccination in achieving the ambitious aim of eradicating malaria over the next five decades. This study proposes a modeling approach using Long-Short Term Memory (LSTM) neurons to forecast the trends of malaria incidence and mortality under three vaccination scenarios until 2070. The two meticulously chosen models, devoid of overfitting or underfitting issues post their training on the data, achieved a noteworthy score (R2) of 91.6% and 86.8% on the test set. Projections for the next half-century indicate that, on the whole, vaccination strategies curtail the malaria burden in the examined countries in the mid-to-long term, compared to scenarios without vaccination. Further analysis reveals that the scenario employing the R21 vaccine exhibits potential in reducing malaria incidence, whereas the RTS vaccine scenario demonstrates effectiveness in minimizing deaths compared to the R21 scenario. However, even in the most optimistic scenario witnessed for Nigeria in the long run, the models don’t foresee complete eradication of malaria by 2070. Consequently, countries must not solely rely on vaccination but must also implement decisive measures to swiftly eliminate the looming threat of malaria. Keywords: LSTM Neural Networks, Malaria dynamics, Prediction.